Ichimoku for Dummies

Shattering the cloud

18 FEBRUARY 2013

Dag Mller +46 8 506 231 29 ICHIMOKU COMPONENTS: TENKAN-SEN (Conversion Line)
(8 day hi+8 day lo)/2

KIJUN-SEN (Base Line)

(21 day hi+21 day lo)/2

SENKOU SPAN A (Leading Span A)

(Tenkan + Kijun)/2 shifted 21 days ahead

SENKOU SPAN B (Leading Span B)

(55 day hi+55 day lo)/2 shifted 21 days ahead

KUMO (Cloud)
Difference between Leading Spans A & B

CHIKOU SPAN (Lagging Span)

Todays close shifted 21 days back

CANDLES
*The suggested 8/21/55/21/21 setting is not the original setup. Traditionally a 9/26/52/26/26 is used .

Period Op/Hi/Lo/Cl

The Ichimoku Kinko Hyo (or Equilibrium at a glance) charting system offers deep insight into market dynamics in any timeframe analysis, from High-frequency trading to a Grand Super-cycle investment guide (though it is mainly used in the daily, weekly & monthly candle perspective). The system consists of several parts which all can be singled out as stand alone tools offering partial insight regarding trends and support/resistance and advice entry & exit strategies. Put together as a concept, the system adds increased accuracy and confidence. The Cloud (or Kumo) shows direction of trend (if any) at-a-glance and the faster Tenkan & Kijun components interact with the slower Kumo for better timing and money management. The lagging Chikou Span confirms trend by comparing the current rate with historical prices. All components are in essence moving averages tweaked in different ways to capture and confirm trends, support & resistance based on activity today and on past volatility and trends which are having impact on present moves. This working paper intends to explain the system building blocks and their respective usage. Examples used here are mainly FX market snapshots, but the system works well in all financial markets such as Equities, Commodities and Fixed Income markets. The Ichimoku system can be combined with other studies such as; Candlestick formations, Price-channels, Slow Stochastics, Fibonacci calculations and Elliott Wave analysis.

You can also find our research materials at our website: www.mb.seb.se. This report is produced by Skandinaviska Enskilda Banken AB (publ) for institutional investors only. Information and opinions contained within this document are given in good faith and are based on sources believed to be reliable, we do not represent that they are accurate or complete. No liability is accepted for any direct or consequential loss resulting from reliance on this document. Changes may be made to opinions or information contained herein without notice.

FX Ringside

The Ichimoku Kinko Hyo charting system

In an attempt to straighten out a few question marks we herein explain the basic concepts of the Ichimoku approach, the thoughts behind its construction and its building blocks. This is a working paper to be completed later with more findings as regard the Ichimoku Kinko Hyo charting system. Later we will, amongst other things, dig deeper into optimization on what parameters are best used in various timeframe analyses and on how they differ on selected markets. higher lows is the basic definition of a bullish trend. This trend could be viewed as an oscillating process around a fair value or equilibrium that changes over time with the trend as the market adjusts to a new higher level. In the described uptrend this equilibrium moves higher over time, while in a downtrend it moves lower. Its worth pointing out that in this concept the equilibrium rate is not determined by any outside parameters thought to affect the charted asset, instead the equilibrium rate in this study is solely set by historical rates and how they have developed over time. Another centrepiece in this approach is the thought that; what we do today will have an impact on future price movements. A strong trend now and high volatility will create stronger future resistance and support levels (since more positions are taken in such a market compared to what is traded in a low volatile range). This thought is what creates the most eyecatching feature of the Ichimoku approach; the Kumo or Cloud which will be described in detail later. Another thought focuses on the importance of checking price action today with earlier prices (to confirm the direction of a trend). The system also detects non-trending conditions or ranges, but is primarily built to signal a neutral stance under such conditions and not designed for a range trading approach.

SHORT HISTORY & BASIC CONCEPTS

The system was developed in Japan during the 1930s by Mr. Goichi Hosoda, but it wasnt released in public until 1968. Ichimoku Kinko Hyo means Equilibrium chart at a glance. To many market participants this study is still categorized as an exotic indicator. Though its truly not one indicator, but rather a set of tools, put together, building a complete charting system. While all of its components may be used as single stand-alone indicators, they work better together as a system. You could say it is a Swiss army knife set of tools, used to identify trends, define levels of support & resistance, generate buy & sell signals (both for entry and exit) as well as identifying stretches. For stretches and exit strategies there are a few other tools outside the Ichimoku system when put together with the Ichimoku system could add accuracy and confidence. The basic observations made by the developer puts focus on the fact that prices or rates on any financial asset or security fluctuates over time, either in ranges when the market is balanced or trending when demand or supply is dominating in the market. Prices trend higher when demand is greater than supply and trend lower when forces reverse. A trend consists of impulses in the underlying direction and corrections against the prevailing trend when impulses reach points where the move gets exhausted. The fine tuned interrelations between short- & longer term interests create waves within waves and this behaviour normally prevents a digital market where prices otherwise would gap from one level to another based on new market information shifting the balance. Over time this process creates trends and in the case of an uptrend as an example; prices are pushed higher (on demand superseding supply). At one point the move becomes exhausted and correctional forces kick in. Once the correction fades the underlying trend takes precedence once again, presumably from a higher price level than from where the trend started. A series of impulses into higher highs and corrections fading at

SYSTEM COMPONENTS
The system rests on five components, or building blocks. All are essentially moving averages tweaked in different ways and designed to move higher or lower with the underlying currents that eventually (and hopefully) create trends. They are constructed to capture shorter and longer oscillations within trends to indicate the direction of the trend(s), to generate buy & sell signals, to identify stretches (where the market is thought to be overbought or oversold) and to provide the trader with potential levels of support and resistance (now and in the future). The system was originally built to capture short-term timeframe market conditions by using daily observations. The developer used 9, 26 & 52 day averages based on the 6 working day/week which was a normal working week when the system was designed. Under these circumstances 9 days = 1 week, 26 days = one month & 52 days = two months. Many claim that this set of parameters still work best, though in a five day working week this setting could however be argued to be altered to 7 , 21 & 42 days. We have so far not historically optimized these

FX Ringside

settings, so at the time of writing we have no statistical proof of what actually works best. But to be true to our belief that there is more than just a self fulfilling factor behind the Fibonacci number sequence: 1-1-2-3-5-8-13-21-34-55-89-144-233 (if you read Dan Browns The Da Vinci Code you know what we talk about), we are evaluating the use of a different setup. For uniform multi-timeframe purposes (intradaily, daily, weekly & monthly analysis) and for the reason that we strongly believe that there is more behind the Fibonacci numbers than just Mumbo Jumbo, we experiment with the alternative 8/21/55 setting (the Fibos closest to the original setting where the first two better describes a modern 1 week and one month). As a comparison, in approaches outside the Ichimoku system, we successfully use moving average settings such as 21-, 55- & 233 days rather than standard 20-, 50- & 200 day moving averages. We will later conduct optimization tests to find out what actually works best in the markets and timeframes we follow.

The shortest measure of Equilibrium in this study is the Tenkan which is based on an eight period measure (nine in the original setup). When it is positively sloped the short-term oscillations within a broader perspective are thought to favor the upside and when spot trades above the Tenkan its thought to act as the first, near-term support level. The opposite rules when it is negatively sloped. But overall trending conditions measured through other components within the system must be considered before making any trade decisions. Breaking a bullishly sloped Tenkan from above is normally not enough to justify a short position since it is merely a warning that the underlying impulse may reverse into a correction.

THE TENKAN AND THE KIJUN

Instead of calculating the average based on for example 21 period closes (for the Kijun) to result in a rolling average of the current last + the recent 20 closes to represent the 21 period equilibrium, the Ichimoku system instead uses: (the highest high) + (the lowest low) over the recent 21 period divided by 2. This calculation is thought to better represent the true equilibrium over the measured period than what a simple moving average does.

A slightly broader perception of Equilibrium is depicted by the Kijun which is based on a 21 period measure (26 periods in the original settings) compared to the Tenkan rolling eight period calculation. The Kijun is slower and normally thought to better show Equilibrium and thus the trend in the timeframe analyzed. It can also be said to be the trend-divider, so when trading above it, the market is crudely thought to be trending higher, and when trading below it, the market is considered to be trending lower.

When comparing a standard simple 21 period moving average (blue) to the 21 period average used within the Ichimoku system (red), called Kijun Sen (or Base Line) it is quite often so that the latter better depicts actual support and resistance than what the former does. This way the average also goes flat when there are no new higher high and/or lower low during the measured period that change the calculation result. This way the average better displays nontrending conditions through a flat equilibrium point rather than a sloping average.

But the slope (if any) of the Kijun is also essential to determine where the undercurrent is flowing and if it is doing a good job doing so. A flat Kijun indicates short-term consolidation and quite often such

FX Ringside

conditions attract a market that has run slightly ahead of itself. Once a short-term correction is completed, closer to the Kijun the underlying move may continue in line with the overall trend in the market and timeframe analyzed.

THE TENKAN/KIJUN CROSS

The faster Tenkan and the slower Kijun conducts repeated crossovers. These shifts are considered as buy or sell signals, depending on whether the Tenkan cross the Kijun from below (buy signal) or if it crosses from above (sell signal). A bullish signal below a positively sloped Cloud is a buy signal, but considered weaker, a bullish crossover inside the Cloud (preferably when it is bullishly slanted) is defined as moderate and a bullish crossover above a positively sloped Cloud is considered strong. The opposite obviously rules in the case of a bearish crossover (over the Kumo or Cloud = weak, inside the Kumo moderate and below it = strong). (We come back to the Kumo or Cloud, its construction and interpretations in just a little while).

The Kumo or the Cloud is the most eye-catching feature within the system and it instantly, or at-aglance tells a trader if the market is trending higher, lower or not trending at all. The Kumo is the area between Senkou Span A and Senkou Span B, shaded in different colours depending on whether the Senkou Span A is above Senkou Span B (bullish) or if Span A is below Span B (bearish). A wider Kumo today means that volatility and trend were stronger 21 periods ago which provide stronger support or resistance today. A narrower (or no) Kumo today means volatility was less 21 periods ago which provides weak or no resistance/support today. The latter creates a better window for the underlying trend to shift direction. The savvy Ichimoku trader may also have insight in possible future (up to 21 periods ahead) shift in the underlying trend depending on what is happening today or just recently happened. A positively sloped Cloud with a flat floor is bullish with the faster Span A indicating a move up from longer-term Equilibrium. The opposite rules for a flat top Kumo with Span A increasing the distance to a sideways Span B. For the purpose to quickly (again: at-a-glance) scan a selected market in a chosen timeframe it is preferred to colour the Kumo differently depending on whether Span A is above or below Span B

THE SENKOU SPAN A & B AND THE KUMO

The Senkou Span A and Senkou Span B are also rolling averages. The Senkou Span A is the Tenkan + the Kijun divided by two to indicate an average of the two short-term workhorse Equilibrium calculations. The Senkou Span B is calculated the same way as the Tenkan & the Kijun but it is based on 55 periods (52 in the original setting). The Senkou Span B is the slowest measure in the system, since it holds the largest number of observations in its calculation). Span A also oscillates around Span B similar to the Tenkan/Kijun oscillation, but less frequently. This also creates indications of shifts in sentiment and the deep underlying trend in the market and timeframe analyzed. Though another thought is added about positions taken today having an impact over future moves, support and resistance levels. For this reason the two Span A & Span B are both shifted 21 periods out in the future (26 periods in the original settings). More volatility/stronger trends today indicates more risks taken while low volatility/no trend does not create the same kind of position footprints.

At occasions when the market is breaking a bearish flat-top Kumo from below, it could be wise to await an initial pullback after the break before entering longs, since the flat Span B depicts a longer term range where the measured Equilibrium is not moving higher nor is it falling over time and during such condition the Equilibrium has a strong attraction pull. So after an initial bullish break, the market tends to fall back to or just below the violated high end of the Kumo before extending the bullish impulse. The trader who waits for the pullback runs a reduced risk of being caught long and wrong in the initial pullback after the break higher. The opposite rules for when bearishly breaking a bullish flatbottomed Kumo.

FX Ringside

THE CHIKOU SPAN (LAGGING SPAN)

The Chikou Span or Lagging Span is exactly what its name depicts. It is simply todays spot shifted 21 periods backwards (the original setting is 26 periods). This line is used to compare todays price with the price 21 periods back. A current spot above the spot 21 periods back implies uptrend and a current spot below the spot 21 periods back indicates downtrend. A current spot at or very close to the spot 21 periods back is neutral, or if current spot is impulsively trending and is just about to cross through the spot rate 21 periods back it would be a bullish confirmation sign. The opposite would rule for a current spot impulsively trending through spot 21 periods ago.

STRETCHES
Any trend is essentially an oscillating process around a fair value. In a trend this fair value or "equilibrium" changes over time. This is true for what ever timeframe is analyzed. Within any timeframe there are different reasons for participation which may be shorter- or longer term oriented. In the short-term or daily candle timeframe there will be those viewing the "Tenkan" as "equilibrium" and there will be those measuring fair value from the "Kijun" or even the "Senkou Spans" and thus the "Kumo". But this must also be true in the medium-term or weekly candle timeframe or the long-term, monthly candle horizon or in any other time frame ranging from a "Grand Super-Cycle" perspective down to the "tic-by-tic" high-frequency trading. In any of those the "Tenkan", "Kijun" and the "Kumo" all represent perception of value depending on which segment within the timeframe the market participant belongs to. A stretch may be defined through the sheer distance between spot and the fair value or "equilibrium" of choice. If overall uptrend indications are in place and spot is considered as stretched compared to the selected measure of "equilibrium" after a correction lower has taken place, a long position should be considered and if the impulse reaches a wide distance to "equilibrium" on the other side and in line with the trend, large enough to make trend following positions likely to be less additionally rewarding, take-profit or at least reduction should be considered. The opposite obviously rules for conditions in a downtrend.

A bullish Chikou Span cross through spot 21 periods ago that occurs above the Kumo is thought to be a stronger signal than if the cross appears in or below the Kumo and the opposite obviously rules for a bearish cross; stronger below the Kumo than if it materializes in or above it. Another practical tool is to draw horizontal support and resistance levels based on the Chikou Span since it is recorded through important closing prices (perhaps less valuable in 24hr markets like FX, while in equities and bonds the closing price is a more important piece of information).

FX Ringside

Another angle may be plausible; if the market is trending strongly, the stretch as measured by the distance to fair value, may be calculated with the slower Kijun in mind, while the stretch during the correction could refer to the distance to the Tenkan instead. Other "non-Ichimoku" tools to measure "stretch" could be the use of volatility sensitive Bollinger Bands or non-volatility sensitive "Price-channels". Bollinger Bands will not be further discussed in this paper. A "Price-channel" could be a simple, - or exponentially weighted Western-style moving average calculated over the "n" recent close prices, shifted up AND down x%. Those shifted moving averages should be designed to capture ~95% of all historic prices within the moving average band. Prices outside this band thus have very poor odds to be additionally extended prior to a healthy correction, thus indicating areas of profit-taking or resistance & support. In for example an uptrend profits should be taken on bearish candles near, at or above the high end of the "Price-channel" and new longs should be considered on bullish candles near, at or below the lower end of the "Pricechannel". This approach may be used in various timeframe analysis, such as; Hourlies (micro-term), Dailies (short-term), Weeklies (medium-term) or Monthlies (long-term). Parameters we use in those are percentage deviation from these averages: 24 hours (=1 global day), 21 days (=1 month), 26 weeks (=-year) & 24 months (=2 years). The percentage deviation from the average to make up the pricechannel differs from market to market depending on its volatility. Up to date we have not optimized the model while they are only eyeballed to contain >95% of all trading in any market in any timeframe. Optimization may add accuracy and confidence and we might do this later.

24 hour Price-channel

21 day Price-channel

26 week Price-channel

24mt Price-channel)

FX Ringside

The disadvantage (and beauty) with Price-channels is the manual handling and consideration of the value of the Price-channel during exceptional market conditions. The first manual override must be considered if/when market volatility conditions permanently change. If the market over time becomes increasingly volatile, the percentage deviation from the average must be increased and if volatility falls markedly over a long period of time, this deviation must be narrowed. A second consideration is that when the market analyzed becomes exposed to a chock, an occurrence which most market participants couldnt be aware of before the information was released. In such cases the stretch parameters as depicted by the boundaries of the Price-channel must be temporary overlooked since the market must be allowed to interpret new information and re-price fair value. The savvy trader knows this and looks elsewhere for possible turning points rather than mechanically acting on what is stipulated by a stretch in respect to the Pricechannel under normal conditions. Yet another consideration must also be made when a market trends steadily and strongly. Under such conditions the market may continue to progress (up or down) despite stretched conditions a lot longer in both price and time than anticipated. The trader should then look for candlestick formations around the outer ends of the Price-channel and not reduce or close a rewarding trend-following trade unless clear candlestick formations strongly indicate that a correction is about to unfold. Another non-Ichimoku tool that works well in combination with the Ichimoku system is the use of slow stochastics. We will herein not go into details how the slow stochastics are calculated, but consider the tool to be able to help a trader to identify; overstretches, price/momentum divergence, trend and are also creating buy- and sell-signals at crossovers. We will focus on the first two, since the Ichimoku system itself generates enough buy- & sell-signals. More would only risk over-information and would be confusing and trends are also well indicated by the Ichimoku system so additional support to detect trends is really not needed. Within any timeframe, there are shorter- or longerterm reasons for participation. For example in the short-term or daily-candle perspective there are those playing it very short-term and those who are less short-term, but still not qualifying as medium-term investors. For this purpose we add more than just one slow-stochastic study in the same chart. The calculations with very few observation periods in the formula will swing faster and more frequently reach overstretched readings while those calculations based on many observations will move slowly and less often indicate a stretch. At occasions all segments within a timeframe converge into a long or short

position, depicted by all Slow-stochastic models becoming highly or lowly grouped. This is when thelast-buyer-has-bought or the-last-seller-has-sold which could translate in a notable turnaround. To be true to our Fibonacci numbers we use: 13, 21, 34, 55, 89, 144 & 233 period calculations for the slow-stochastics. Put together in one chart we call them multi-stochastics. It is possible to work these out with a good result in all timeframes (hourlies, dailies, weeklies and monthlies). The multi-stochastic model shows more than just stretches. At occasions when a momentous move in price creates what looks like a stretch in the multistochastics a correction should be expected, but quite so often additional highs are scored after a brief correction. Though this time around the multistochastics may not be able to score the same kind of stretch the second time. This creates a divergence between the move in price and the momentum indicator. If this happens after a rally, it would be called a bearish divergence and if it this unfolds after bearish rout it would be called a bullish divergence. The divergence normally kicks in late in a trend, warning of its termination and following reaction in the opposite direction. This is another way to depict what an Elliott wave analyst would call the fifth and final wave in a trending impulse which frequently takes the shape of wedge.

EXIT STRATEGIES
While the Ichimoku system offers a plethora of approaches for both entry- and exit strategies, the use of a trailing-stop could add confidence. A Parabolic SAR or a 5period low/5period high rolling average could also be added to the system. We will here focus on the latter. In a short-term or daily candle timeframe uptrend, a long position should be entered and as a trailing stop the average 5day low may be used as a stop profit. Our experience is that the stop in this case only should be triggered on a session CLOSE below the average 5day low (or a CLOSE above the average 5day high if the trailing stop protects a short position) and disregard intraday violation. This obviously leaves substantial intraday risk on the table so this tactics

FX Ringside

must have a safety net attached through a maximum intraday stop in the case the market is in a runaway mode and closes the day far away from the actual trigger level.

ADDITIONAL CONSIDERATIONS
Trailing stop at the average 5day high

In combination with what is described in this paper, candlestick and volume analysis should used topped with Fibonacci, retracement and projection calculations and a prudent Elliott wave approach.

Trailing stop at the average 5day low

The average 5day low/5day high is dependent on todays low + the four recent lows and todays high + the four prior highs. Todays extremes thus make up one-fifth of the calculation so it could be worth considering to push the model one day ahead to avoid fluctuation in its value depending on where todays extremes actually fall into place (since yesterdays 5day high/5day low then would be used for todays trailing stop).

Put the pieces together and it could look like this

Dag Mller